Some terms: KAB = Knowledge Attitude Behaviour KAP = Knowledge Attitude Practice KPC = Knowledge Practice Coverage These are related approaches. We can say that KPC is KAP or KAB with an explicit coverage component. The terms relate to a general approach rather than a standard approach. I think there are five main issues to consider. These are sampling methods, sample size, analysis, question design, and validity / utility of findings. Sampling : KAP / KPC surveys should use a sample that is representative of the population they purports to represent. For a large population such as mothers of children aged under two months sampling methods such as those used in EPI surveys (SMART is an EPI design) can be used. For more restricted populations you may have to use a chain-referral sampling method. An example of this is coverage of CMAM services for SAM cases using the CSAS or SLEAC method. These methods use active and adaptive case-finding (a type of chain referral sample) and apply a short KAP type questionnaire to elicit barriers to coverage. Sample size : These types of survey are little different from other questionnaire surveys in this regard. Standard sample sizes calculation formulae can be used. Complications are multiple indicators and small populations. In the case of multiple indicators you should list your indicators, make a guess their values, specify a desired precision, and calculate a sample size for each. The maximum sample size from all these calculations is the required sample size. Usually we repeat this process a few times and compromise on desired precisions to arrive at a feasible sample size. As a rule of thumb you should plan for a sample size no smaller than for an EPI survey (i.e. 30 clusters of 7). If you have a small population (e.g. SAM cases) then you can make an estimate of population size and adjust sample sizes using a finite population correction. If you have previous survey data (or other relevant data) you may be able to make sample size savings by adopting a Bayesian approach as is done in SQUEAC stage III surveys. Analysis : Be sure to analyse your data using methods that account for the sample design. Assuming a simple random sample will likely give spuriously precise estimates. Question design : The main issue here is to avoid questions that restrict answers to what you think you know and that prompt the respondent to reflect back program messages or just positive messages (in the UK's National Health Service KAP survey questionnaires are often referred to as "happy sheets" as they tend to tell us what we want to hear). This means that many questions will be open-ended or have "Other ... specify ..." options. Multi-variable scales can be useful as data can be checked for internal consistency. Questions should also allow for multiple answers. Use of local languages, terms, and aetiologies is essential. You should (at least) have a qualitative development stage to develop and test questions. You also have to be careful with regard to the context of interviews. When I use KAP type surveys / survey components I tend to keep the questionnaire very focussed and very short. For example, the KAP component of CSAS, SLEAC, and stage III SQUEAC surveys is limited to six questions. Validity / utility : These types of survey are among the most difficult to get right. There have been spectacular failures. At one point the WHO recommended that KAP surveys should no longer be done and published material on replacement techniques. KAP survey are, however, still with us. It is important that you realise their strengths and weaknesses. [url=http://www.anthropologymatters.com/index.php?journal=anth_matters&page=article&op=viewArticle&path%5B%5D=31&path%5B%5D=53]Here[/url] is a narrative article with some useful references that outline several issues with KAP style surveys. My personal view is that KAP surveys can provide useful data. The KAP data from CSAS and SLEAC coverage surveys, for example, has proved very useful for designing and reforming CMAM programs and has helped improved program coverage. It is important to realise that KAP data has limitations. KAP data (in CSAS and SLEAC) is very thin compared to what can be discovered by semi-quantitative methods (such as SQUEAC). To summarise ... use with caution. I hope this of some use.